
507Pricing Multiple Lines of Business and Risks
The t copula case is only marginally different from the Gaussian case explained in Box
28.10 and the comments in Box 28.11 for the t copula case will therefore be much sparser,
focussing on the aspects that are different.
3. Put Σ in diagonal form, obtaining D = diag(λ
1
,…λ
n
) (that is, a matrix for which
all elements except those in the diagonal are zero, and the elements in the diagonal are
λ
1
, … λ
n
).
Since Σ is a symmetric matrix, this can always be done. The relationship between
Σ and D can be written as D = V
−1
ΣV where V is the matrix, which has the eigenvec-
tors of Σ as its columns. This relationship ...